MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol. 613: 197–210, 2019https://doi.org/10.3354/meps12894

Published March 21

1. INTRODUCTION

Anthropogenic activities may influence the dietaryniche width of wild populations by modifying theavailability of resources (Van Valen 1965, Newsomeet al. 2015). While human activities often result in re -

source scarcity, sometimes they generate new feed-ing opportunities for species (Votier et al. 2004,Woodroffe et al. 2005, Jennings et al. 2009). This isthe case when predators feed on a resource that iseither produced, raised or captured by humans(Woodroffe et al. 2005). This behaviour, defined as

© The authors 2019. Open Access under Creative Commons byAttribution Licence. Use, distribution and reproduction are un -restricted. Authors and original publication must be credited.

Publisher: Inter-Research · www.int-res.com

*Corresponding author: p.tixier@deakin.edu.au

Importance of toothfish in the diet of generalistsubantarctic killer whales: implications for fisheries

interactions

Paul Tixier1,2,*, Joan Giménez3, Ryan R. Reisinger2,4, Paula Méndez-Fernandez5, John P. Y. Arnould1, Yves Cherel2, Christophe Guinet2

1School of Life and Environmental Sciences (Burwood campus), Deakin University, Geelong, Victoria 3125, Australia2Centre d’Etudes Biologiques de Chizé (CEBC), UMR 7372 Université de la Rochelle-CNRS, 79360 Villiers-en-Bois, France

3Institut de Ciències del Mar (ICM-CSIC), Passeig Marítim de la Barceloneta 27-49, 08003 Barcelona, Spain4CESAB-FRB, Bâtiment Henri Poincaré, Domain du Petit Arbois, Ave Louis Philibert, 13100 Aix-en-Provence, France

5Observatoire PELAGIS, UMS 3462 du CNRS, Pôle Analytique, Université de la Rochelle, 5 allées de l’Océan, 17000 La Rochelle, France

ABSTRACT: Fisheries may generate new feeding opportunities for marine predators, whichswitch foraging behaviour to depredation when they feed on fish directly from fishing gear. How-ever, the role of diet in the propensity of individuals to depredate and whether the depredatedresource is artificial or part of the natural diet of individuals is often unclear. Using stable isotopes,this study investigated the importance of the commercially exploited Patagonian toothfish Dissos-tichus eleginoides in the diet of generalist subantarctic killer whales Orcinus orca depredatingthis fish at Crozet (45°S, 50°E). The isotopic niche of these killer whales was large and overlappedwith that of sperm whales Physeter macrocephalus from the same region, which feed on toothfishboth naturally and through depredation. There was no isotopic difference between killer whalesthat depredated toothfish and those that did not. Isotopic mixing models indicated that preygroups including large/medium sized toothfish and elephant seal Mirounga leonina pups repre-sented ~60% of the diet relative to prey groups including penguins, baleen whales and coastalfish. These results indicate that toothfish are an important natural prey item of Crozet killerwhales and that switching to depredation primarily occurs when fisheries facilitate access to thatresource. This study suggests that toothfish, as a commercial species, may also have a key role asprey for top predators in subantarctic ecosystems. Therefore, assessing the extent to which predators use that resource naturally or from fisheries is now needed to improve both fish stockmanagement and species conservation strategies.

KEY WORDS: Diet · Fisheries · Southern Ocean · Killer whale · Stable isotopes · Fishery interactions

OPENPEN ACCESSCCESS

Mar Ecol Prog Ser 613: 197–210, 2019

‘depredation’, has been increasingly reported bothin terrestrial (e.g. predators feeding on livestock;Sillero-Zubiri et al. 2007) and marine environments(e.g. predators feeding on fish raised in farms orcaught in fishing gears; Northridge & Hofman 1999,Gilman et al. 2007, 2008, Read 2008).

The depredated resource may already be part ofthe natural diet of predators, with access facilitatedby humans, or it may be an entirely artificial re -source which would not otherwise have been usedby predators in natural conditions. This distinction iscritical in understanding the underlying behaviouralmechanisms of predators switching from a naturalto a depredated resource (Boitani & Powell 2012).Generalist predators may be more likely to depre-date artificial new resources opportunistically whilehighly specialised predators may depredate a re -source only if it is already part of their natural diet(Stoddart et al. 2001, Sidorovich et al. 2003). Know-ing the importance of the depredated resource inthe natural diet of predators is also needed to assessthe effects of depredation and fisheries on wild pop-ulations, fish stocks and ecosystems as a whole.Depredation may substantially modify the energybalance of the predator and its role in ecosystemfood web dynamics (Woodroffe et al. 2005). Forinstance, if the depredated resource is fully artificialfor the predator, depredation may lead to decreasedavailability of that resource for other functionalgroups in the ecosystem, subsequently affectingthese groups through trophic effects (Woodroffe etal. 2005). Also, by feeding on fish caught in fishinggear, marine predators may cause increased anddifficult-to-quantify mortality for fish stocks, therebyincreasing the catches needed for fisheries to reachtheir quotas (Gilman et al. 2013, Gasco et al. 2015,Mitchell et al. 2018).

In subantarctic waters, extensive commercial long-line fisheries target economically highly valuablePatagonian toothfish Dissostichus eleginoides (here-after ‘toothfish’). These fisheries provide artificialfeeding opportunities for a range of large marinepredators through discards (primarily for albatrossesand petrels) and depredation (for odonto cetes)(Tasker et al. 2000, Kock 2001, Kock et al. 2006).Toothfish are large (0.5−2 m) fish that dominate thebiomass of the bathypelagic zone, but may also befound in the meso- and epi-pelagic zones (Collins etal. 2010). However, the role of toothfish in subantarc-tic ecosystems and their importance as a natural preyfor predators is unclear (Cherel et al. 2000, 2017,Constable et al. 2000). Specifically, determining theextent to which predators naturally feed and rely on

toothfish is critical to assess the impacts of exploita-tion of that resource by fisheries on the conservationof subantarctic predators, many of which are threat-ened or endangered (Croxall et al. 2012).

Killer whales Orcinus orca are one of the main spe-cies depredating toothfish from subantarctic longlinefisheries (Kock et al. 2006). Unlike some other re -gions, where killer whales have highly specialisedprey preferences (Similä et al. 1996, Ford et al. 1998,Foote et al. 2009), subantarctic populations have rel-atively broad dietary niches that include mammals,birds, fish and sometimes cephalopods (Guinet &Jouventin 1990, Guinet 1992, Guinet et al. 2000, deBruyn et al. 2013, Capella et al. 2014, Reisinger et al.2016, Travers et al. 2018). This generalist diet may bedriven by the subantarctic ecosystem’s spatio-tempo-ral heterogeneity in the availability of high-qualityresources, such as seals, penguins and whales (Laws1977, Knox 2006, Reisinger et al. 2018). This hetero-geneity may force killer whales to supplement theirprimary diet with other prey such as fish and ce -phalopods. While toothfish are a confirmed depre-dated resource, there is no direct evidence of naturalpredation by killer whales on this fish species.

The killer whale population of the Crozet Islands(subantarctic islands located at 45°S, 50°E), hereafter‘Crozet killer whales’, is among the populations thathave most extensively depredated toothfish fromfisheries since the mid-1990s (Roche et al. 2007, Tix-ier et al. 2010, 2016, Guinet et al. 2015). Crozet killerwhales prey on seals, penguins, baleen whales andsmall notothenioids in inshore waters (Guinet 1992,Guinet et al. 2000); not all individuals have switchedto depredation on toothfish from fisheries (Tixier etal. 2015, 2017). Elucidating the extent to which thesekiller whales naturally rely on toothfish as a resourcewhose availability is modified by fisheries wouldtherefore provide insights into the ecological mecha-nisms of prey switching to depredation. Critically,this information would clarify the role of these preyand predator species in subantarctic ecosystem foodweb dynamics and the impacts of fisheries on theconservation of predator populations and fish stocks.Therefore, using stable isotope and diet reconstruc-tion analyses for Crozet killer whales, the aims of thisstudy were to (1) assess the importance of toothfishrelative to other prey items for Crozet killer whales,and compared with other Southern Ocean killerwhale populations and odontocete species, and(2) examine variation in the dietary importance oftoothfish across individuals of the same population,with respect to whether or not they depredated fromfisheries.

198

Tixier et al.: Toothfish in subantarctic killer whale diet

2. MATERIALS AND METHODS

2.1. Species and sample collection

Killer whale skin samples were collected at Crozetfrom biopsies performed remotely on free-rangingindividuals, using a Barnett Rhino 150 lb compoundcrossbow and custom-built darts (Ceta-Dart) equippedwith sterilized stainless steel tips (35 mm length,7 mm diameter). Sampling was conducted oppor-tunistically and passively (i.e. animals were notactively approached or followed using motorizedmeans) from land on Possession Island and from atoothfish fishing vessel operating in the CrozetExclusive Economic Zone (EEZ), from February 2011to December 2012. Samples were collected from themid-lateral region of the body, below the dorsal fin.Sub-samples containing skin tissue were stored in70% ethanol. Only weaned individuals (>2 yr old)were sampled. Sampling occurred only when a sin-gle individual surfaced within 15 m of the sampler,and only when this individual was positively identi-fied by eye during surfacing events directly preced-ing sampling. The sequence made of multiple surfac-ing events and the sampling event was monitored byphotographs and/or video. Photographs of the dorsalfin of the sampled individual were systematicallytaken using a DSLR camera with 400 mm telephotolens, and were used to confirm the identity of thatindividual after sampling, using an existing photo-identification database (Tixier et al. 2014). An addi-tional skin sample was obtained from an individualfound stranded and dead on Possession Island on 17August 2006. The individual was an apparentlyhealthy sub-adult male (total length [TL]: 6.90 m)which was known to be part of the Crozet killerwhale population based on photo-identificationrecords. However, because this sample was collected5 yr before the biopsy samples, at a different time ofyear (winter) and lacked information about the be -haviour of the individual over the pre-sampling pe -riod, isotopic information for this sample was notincluded in the analyses.

Isotopic information from skin samples collectedfrom the Crozet killer whales was first examinedthrough large-scale comparisons with other isotopicinformation available for other killer whale popula-tions and other large odontocete species with differ-ent feeding ecologies and/or different habitats in theSouthern Ocean (south of the Subtropical Front,~40°S). Published isotopic values for weaned killerwhales (>2 yr old) were obtained for one other sub-antarctic (Marion Island, 1000 km west of Crozet at a

similar latitude) and 3 Antarctic populations (TypesB1 and B2 around the Antarctic Peninsula [Durban etal. 2017] and Type C in the Ross Sea [Pitman & Ensor2003]). The isotopic niche of the Crozet killer whaleswas expected to be similar to that of killer whales atMarion, as the 2 populations share similar habitatsand appear to have a generalist feeding strategybased on consumption of the same prey species(Reisinger et al. 2016). In contrast, Antarctic killerwhales, which use a different habitat and specialiseon either fish (Type C; Pitman & Ensor 2003, Krahn etal. 2008), krill consumers (e.g. pygoscelid penguins)(Type B2; Pitman & Durban 2010) or predators of krillconsumers, such as Weddell seals Leptonychotesweddellii (Type B1; Pitman & Durban 2012, Durbanet al. 2017), were ex pected to have limited isotopicoverlap with the Crozet killer whales. Isotopic infor-mation was also compared with that of sperm whalesPhyseter macrocephalus, which also depredate tooth -fish on longline fisheries, and southern long-finnedpilot whales Globi cephala melas edwardii from pop-ulations using similar habitats to that of the Crozetkiller whales at Crozet and/or adjacent waters.Sperm whale skin samples were collected from fish-ing vessels operating in the Crozet and KerguelenEEZs in January and February 2011, using the sameequipment and protocols, and were treated using thesame process as for the Crozet killer whale samples.For southern long-finned pilot whales, whose pelagichabitat overlaps with areas where toothfish fisheriesoperate but were never observed depredating tooth-fish on longlines, published skin isotopic values fromweaned individuals (>3 m in length) at Kerguelenwere used (Fontaine et al. 2015).

Species were considered confirmed prey for theCrozet killer whales if predation was directly ob -served and/or remains were found in the stomachcontents of the individual found dead in 2006(Table S1 in Supplement 1 at www. int- res. com/articles/ suppl/ m613 p197 _ supp. pdf). From these data,species confirmed as prey items and for whichspring/summer isotopic values were available for thestudy at Crozet included southern elephant sealsMirounga leonina (adult females and pups, con-firmed as prey from observations and stomach contents); Antarctic and subantarctic fur seals Arcto-cephalus gazella and A. tropicalis (pups, from stom-ach contents); king penguins Aptenodytes patagoni-cus, gentoo penguins Pygoscelis papua, macaronipenguins Eudyptes chrysolophus, rockhopper pen-guins E. chrysocome filholi (all adults, from observa-tions and stomach contents); and Patagonian tooth-fish (TL 81−174 cm, from observations) (Table 1).

199

Mar Ecol Prog Ser 613: 197–210, 2019

Additional Patagonian toothfish samples for smalland medium size individuals (TL 38−71 cm) were col-lected at Kerguelen (Table 1). For southern rightwhales Eubalaena australis (confirmed as prey fromobservations), isotopic values were from individualssampled near subantarctic Campbell Island (Torreset al. 2017) (Table 1).

2.2. Stable isotope analyses

Killer whale skin samples were first oven-dried at50°C for 48 h to allow ethanol evaporation, thenground and freeze-dried. As stable isotope valuesmay be influenced by the lipid content of the tissue(Lesage et al. 2010, Giménez et al. 2017), 2 successive

200

Species Code Site n Tissue δ13C δ15N δ13C (‰) δ15N (‰) Source(‰) (‰) (adjusted) (adjusted)

Group A −19.5 ± 0.6 8.2 ± 0.5Gentoo penguin (spring) Pygoscelis papua GP_sp CR 11 RBC −18.6 ± 0.3 8.2 ± 0.6 −18.2 ± 0.3 8.3 ± 0.6 This study

Macaroni penguin (spring) Eudyptes chrysolophus MP_sp CR 10 RBC −19.4 ± 1.0 7.0 ± 0.2 −19.0 ± 1.0 7.1 ± 0.2 This study

Macaroni penguin (summer) Eudyptes chrysolophus MP_su CR 20 RBC −20.0 ± 0.7 8.6 ± 0.4 −19.6 ± 0.7 8.7 ± 0.4 This study

Rockhopper penguin (spring) Eudyptes chrysocome filholi RP_sp CR 10 RBC −20.2 ± 0.4 8.1 ± 0.5 −19.8 ± 0.4 8.2 ± 0.5 This study

Rockhopper penguin (summer) Eudyptes chrysocome filholi RP_su CR 10 RBC −20.8 ± 0.3 8.6 ± 0.4 −20.4 ± 0.3 8.7 ± 0.4 This study

Southern right whale Eubalaena australis SRW NZ 18 Skin −19.8 ± 0.6 8.1 ± 0.7 −19.8 ± 0.6 8.2 ± 0.7 Torres et al. (2017)

Group B −21.4 ± 0.4 10.4 ± 0.3Southern elephant seal (adult females)Mirounga leonina SES_fem CR 70 WB −21.9 ± 0.7 10.2 ± 0.2 −22.3 ± 0.7 10.9 ± 0.17 This study

King penguin (spring) Aptenodytes patagonicus KP_sp CR 11 RBC −21.8 ± 0.4 10.1 ± 0.2 −21.4 ± 0.4 10.2 ± 0.2 This study

King penguin (summer) Aptenodytes patagonicus KP_su CR 10 RBC −22.2 ± 0.2 10.3 ± 0.2 −21.8 ± 0.2 10.4 ± 0.2 This study

Patagonian toothfish (TL 43 cm) Dissostichus eleginoides TOP_43 KE 6 Muscle −20.3 ± 0.4 10.0 ± 0.8 −20.3 ± 0.4 10.0 ± 0.8 This study

Group C −20.5 ± 0.5 12.7 ± 0.4Southern elephant seal (pups) Mirounga leonina SES_pup CR 70 WB −21.6 ± 0.7 11.6 ± 0.4 −22.0 ± 0.7 12.3 ± 0.4 This study

Antarctic fur seals (pups) Arctocephalus gazella AFS_pup CR 10 WB −20.7 ± 0.5 12.4 ± 0.2 −21.1 ± 0.6 13.1 ± 0.2 Cherel et al. (2015)

Subantarctic fur seals (pups) Arctocephalus tropicalis SAFS_pup CR 10 WB −19.4 ± 0.2 12.0 ± 0.3 −19.8 ± 0.2 12.7 ± 0.3 Cherel et al. (2015)

Patagonian toothfish (TL 63 cm) Dissostichus eleginoides TOP_63 KE 17 Muscle −19.2 ± 0.6 12.6 ± 0.9 −19.2 ± 0.6 12.6 ± 0.9 This study

Group D −18.9 ± 0.6 14.4 ± 0.6Patagonian toothfish (TL 95 cm) Dissostichus eleginoides TOP_95 CR 14 Muscle −18.3 ± 0.5 14.2 ± 0.7 −18.3 ± 0.5 14.2 ± 0.7 This study

Patagonian toothfish (TL 107 cm) Dissostichus eleginoides TOP_107 CR 10 Muscle −19.1 ± 0.5 14.3 ± 0.4 −19.1 ± 0.5 14.3 ± 0.4 This study

Patagonian toothfish (TL 160 cm) Dissostichus eleginoides TOP_160 CR 22 Muscle −19.3 ± 0.8 14.9 ± 0.7 −19.3 ± 0.8 14.9 ± 0.7 This study

Table 1. Mean ± SD δ13C and δ15N values for prey species and prey groups of the Crozet killer whale used in the MixSIAR models. Preygroups were determined via Ward’s hierarchical clustering based on isotopic similarities. Prey species were differentiated based on age class(adults vs. juveniles; here, ‘pups’ for seal species), sex, season in which sampling occurred or size (total length [TL] for fish species) when rel-evant, and an abbreviation code is assigned to each. Sampling sites included Crozet Islands (CR), Kerguelen Islands (KE) and New Zealand(NZ). Isotopic values are provided for the type of tissue sampled: red blood cells (RBC), whole blood (WB), skin or muscle. These values were

adjusted to represent muscle values for all prey species

Tixier et al.: Toothfish in subantarctic killer whale diet

and identical lipid extractions were conducted usingcyclohexane, each through 1 h sonication and subse-quent centrifugation at 3000 rpm (i.e. 1613 × g). Thelipid-extracted samples were then oven-dried againat 50°C for 48 h before being sub-sampled down to0.3−0.4 mg. These sub-samples were processed at LIttoral ENvironnement et Sociétés (LIENSs; Univer-sity of La Rochelle, France) through a continuous-flowisotope-ratio mass spectrometer (Micromass Isoprime)paired with an elemental analyser (Euro Vector EA3024) for carbon and nitrogen isotope relative abun-dance (13C/12C and 15N/14N, respectively). The isotopiccompositions are reported in the conventional δ nota-tion as the per mil (‰) deviation relative to the stan-dards Vienna Peedee Belemnite (for carbon) and air(for nitrogen), expressed in parts per thousand (‰).Within-run (n = 10) replicate measurements of in -ternal laboratory standards (acetanilide) indicatedmeasurement errors <0.15‰ for both δ13C and δ15Nvalues. Samples with a C:N mass ratio <3.6 were con-sidered lipid-free (Yurkowski et al. 2015, Giménez etal. 2017) and included in subsequent analyses. Preysamples for which the analysed tissue was muscle(toothfish) were processed using the same protocol asfor killer whale skin, including cyclohexane lipid ex-traction. For the other prey (elephant seals, fur sealsand penguins), where isotopic values were measuredfor whole blood or red blood cells, values were ad-justed to represent muscle using adjustment valuespresented in Reisinger et al. (2016). The isotopicmethod was validated in the southern Indian Ocean(encompassing the killer whale feeding areas), withδ13C values of consumers indicating their foraginghabitats (Cherel & Hobson 2007) and their δ15N valuesincreasing with trophic level (Che rel et al. 2010).

The isotopic niche width of the Crozet killer whalepopulation was estimated and compared to otherpopulations of killer whales, sperm whales andsouthern pilot whales within the Southern Ocean in aBayesian framework using multivariate ellipse-basedmetrics (Jackson et al. 2011). Standard ellipse areascorrected for sample size (SEAc) and Bayesian Stan-dard Ellipse Areas (SEAB) were calculated for eachgroup. SEAB was estimated using 105 posterior drawsand used to statistically compare niche metrics,which included niche width and niche overlap be -tween groups. The niche overlap for 2 given groupswas calculated as an isotopic area of overlap from themaximum likelihood fitted ellipses of the 2 groups(Jackson et al. 2011). All niche metric calculationsand comparisons were conducted with the package‘SIBER’ (Jackson et al. 2011) in R v.3.4.1 (R Develop-ment Core Team 2017).

The effect of toothfish depredation on the isotopicvalues of Crozet killer whales was examined througha 2-state index assigned to each sample. Sampleswere categorised as ‘depredating’ or ‘non-depredat-ing’ depending on whether the biopsied individualswere observed interacting with fisheries in the 24 or48 d preceding the sampling date. These 2 periodswere defined based on isotopic half-time turnoverrates estimated for bottlenose dolphin Tursiops trun-catus skin: 24 ± 8 d for carbon and 48 ± 19 d for ni -trogen (Giménez et al. 2016). Thus, depredating samples were samples from individuals that werephoto graphed at least once while depredating tooth-fish caught by fishing vessels during the 24 d beforesampling for δ13C analyses, and 48 d before samplingfor δ15N analyses. Photographs were taken from fish-ing vessels by fishery observers, who are presentonboard licenced toothfish longliners for all fishingtrips. They monitor 100% of the fishing operationsand provide a quasi-systematic (on average >95% offishing days with killer whale presence around ves-sels covered) photo-identification effort during killerwhale−fishing gear interaction events using DSLRcameras with 400 mm telephoto lenses. Non-depre-dating samples were samples from individuals forwhich the biopsy was performed when no fishingoccurred in the Crozet EEZ 24 or 48 d before sam-pling, based on the PECHEKER database (Martin &Pruvost 2007), or from individuals that were notphoto graphed from fishing vessels 24 or 48 d beforesampling. The absence of fishing vessels operatingillegally inside and/or in the vicinity of the CrozetEEZ during the 48 d preceding biopsies was checkedthrough satellite and ship-based surveillance datarequested from the French administration. This ruledout the possibility that killer whales had depredatedtoothfish from vessels other than the ones from whichwe received data. Additionally, recent satellite/diverecorder data from a killer whale depredating tooth-fish at South Georgia indicated that depredationevents only occurred in the vicinity of the vessel dur-ing gear retrieval phases, which is when observersprovide photo-identification effort (Towers et al.2019).

Niche metric comparisons and statistical tests per-formed on δ13C and δ15N values were used to assessdietary variations between depredating and non-depredating samples. SEAc and SEAB were calcu-lated separately for depredating and non-depredat-ing samples; here, this assignment was made usinginformation on the occurrence of depredation in the48 d preceding sampling. After the normality of theδ13C and δ15N values was tested (Shapiro-Wilk test),

201

Mar Ecol Prog Ser 613: 197–210, 2019

differences between depredating and non-depre -dating groups were tested using either parametric (t-test) or non-parametric (Mann-Whitney U-test) tests.

2.3. Diet reconstruction

The relative contribution of various prey items tothe diet of the Crozet killer whales, and the influenceof fishery interactions on this contribution, were as -sessed through Bayesian stable isotope mixing models fitted in the ‘MixSIAR’ package (Stock &Semmens 2013, Stock et al. 2018) in R v.3.4.1 (R De -velopment Core Team 2017). This analysis was con-ducted using prey data only for confirmed preyitems, as described above. The mean stable isotopevalues of these species were used to a priori identifystatistically different clusters through a Ward’s hier-archical cluster analysis (‘hclust’ function in R pack-age ‘stats’) and ANOVAs. MixSIAR models were fit-ted using the individual isotopic values of the Crozetkiller whales (consumer), the mean ± SD isotopic val-ues of prey clusters (sources) and the diet-to-tissuediscrimination factors (DTDF) estimated by Giménezet al. (2016) for bottlenose dolphin skin (DTDF forδ13C = 1.01 ± 0.37‰; δ15N = 1.57 ± 0.52‰). Fisheryinteraction was incorporated in the MixSIAR modelsas a fixed effect using samples categorised as depre-dating or non-depredating based on the 48 d preced-ing sampling. The effect of depredating or non-depredating on the relative contribution of preygroups to killer whale diet was tested through modelselection based on the leave-one-out information cri-terion (LOOic) (Vehtari et al. 2017, Stock et al. 2018).

Models were run with a generalist type prior, 3 Mar -kov chain Monte Carlo (MCMC) chains of 300 000draws and a burn-in of 200 000 draws. The conver-gence of models was checked using both Gelman-Rubin and Geweke diagnostics. Model evaluationand validation were conducted by determining thelikelihood of prey groups being included in the mix-ing polygon of the Crozet killer whales, based onsimulations developed by Smith et al. (2013). Unlessotherwise stated, data are presented as mean ± SD.

3. RESULTS

Biopsy samples were obtained from 18 individualsfrom the Crozet killer whale population (Table S2 inSupplement 1). Values of δ13C of lipid-extracted skinranged from −19.6 to −18.0‰ and values of δ15N from12.5 to 14.3‰ (Table 2). Tests for potential age- andsex-effects on δ13C and δ15N values indicated no sig-nificant differences between males (n = 3) and fe-males (n = 15), nor between adults (n = 14) and sub-adults (n = 4) (t-tests, all with p > 0.5). Similarly, therewere no significant differences between months ofsampling (n = 9 in February, n = 6 in November andn = 3 in December). The isotopic values of the skin sam-ple collected from the individual found dead in 2006were δ13C = −18.8‰ and δ15N = 13.4‰ (Table S2).

3.1. Stable isotope analyses

The isotopic niche area of the Crozet killer whales,which was estimated at SEAc = 0.64‰2 and SEAB =

202

n δ13C (‰) δ15N (‰) SEAc SEAB Pr % SEAB Source(‰2) (‰2) <SEAB overlap

Orcinus orcaCrozet Islands 18 −19.0 ± 0.5 13.6 ± 0.4 0.64 0.57 − − This studyMarion Island 32 −18.6 ± 0.4 12.3 ± 0.6 0.94 0.87 0.90 3 Reisinger et al. (2016)Ross Sea (Type C) 27 −23.8 ± 0.4 13.2 ± 0.4 0.21 0.2 0.00 0 Krahn et al. (2008)Antarctic Peninsula (Type B1) 11 −22.4 ± 0.4 12.2 ± 0.4 0.45 0.38 0.17 0 Durban et al. (2017)Antarctic Peninsula (Type B2) 8 −22.8 ± 0.3 11.3 ± 0.2 0.20 0.16 0.01 0 Durban et al. (2017)

Physeter macrocephalusCrozet Islands/Kerguelen Islands 6 −18.6 ± 0.4 14.0 ± 0.4 0.70 0.51 0.49 24 This study

Globicephala melas edwardiiKerguelen Islands 65 −18.4 ± 0.5 12.2 ± 0.3 0.36 0.40 0.07 0 Fontaine et al. (2015)

Table 2. Mean ± SD δ13C and δ15N values of lipid-extracted skin and isotopic niche metrics for killer whales Orcinus orca,sperm whales Physeter macrocephalus and southern pilot whales Globicephala melas erwardii within the Southern Ocean.Niche metrics include the standard ellipse areas corrected for sample size (SEAc) and the Bayesian SEA (SEAB). SEAB wasused to estimate the probability (Pr) of the Crozet killer whale isotopic niche being smaller than that of other groups

(Pr <SEAB), as well as the degree of overlap of isotopic niches (% SEAB overlap)

Tixier et al.: Toothfish in subantarctic killer whale diet

0.57‰2, had a high probability of being smaller thanthat of the Marion killer whales (SEAc = 0.94‰2 andSEAB = 0.87‰2), but was likely larger than those ofType B1, B2 and C killer whales sampled in Antarc-tica, and southern pilot whales from Kerguelen wa -ters (Table 2, Fig. 1). However, with a probability of0.49, the isotopic niche width of the Crozet killerwhales was statistically similar to that of spermwhales sampled in Crozet and Kerguelen waters.Niche overlap of the Crozet killer whales was zerowith Antarctic killer whales and Kerguelen southernpilot whales, low with the Marion killer whales (3%)and highest with the Crozet and Kerguelen spermwhales (24%).

In total, 9 of the sampled killer whales were sighteddepredating toothfish from fisheries during the 24 dpreceding sampling, and these samples were thusconsidered as depredating for δ13C comparisons (Ta-bles S2 & S4 in Supplement 1). Values of δ13C of thesesamples (δ13C = −19.0 ± 0.7‰, n = 9) were not signifi-cantly different (Mann-Whitney U-test, p = 0.57)from those of non-depredating samples (δ13C = −19.0

± 0.3‰, n = 9). Values of δ15N of samples from 11 in-dividuals sighted depredating from fisheries in the48 d preceding sampling (δ15N = 13.4 ± 0.4‰, n = 11)were not statistically different from those of samplesfrom individuals that did not depredate from fisheriesover that period (δ15N = 13.8 ± 0.4‰, n = 7; t-test, p =0.09). The isotopic niche area of non-depredatingsamples (SEAc = 0.27 and SEAB = 0.16‰2) was likelysmaller than that of depredating samples (SEAc =0.77 and SEAB = 0.71‰2) (Fig. S1 in Supplement 1).

3.2. Diet reconstruction

In total, 4 statistically different prey groups wereidentified from the Ward’s hierarchical clustering ofδ13C and δ15N values (Figs. 2, 3a & Fig. S2, Table S5in Supplement 1). Group A included species with thelowest δ15N values (8.2 ± 0.5‰): 3 species of pen-guins (Eudyptes spp. and gentoo penguins) andsouthern right whales (Table 1). Group B had ahigher mean δ15N value (δ15N = 10.4 ± 0.4‰) and

203

Fig. 1. Sample-size corrected standard ellipse areas (SEAc; solid lines) and convex hull areas (dotted lines) for killer whales(KIW), sperm whales (SPW) and southern pilot whales (PIW) in Antarctic and subantarctic waters. Individual values of δ13Cand δ15N of skin samples (points) for Crozet (this study), Marion (Reisinger et al. 2016), Antarctic Type C (Krahn et al. 2008),Ant arctic Type B1 and B2 (Durban et al. 2017) killer whales, sperm whales from Crozet/Kerguelen (this study) and southern

pilot whales from Kerguelen (Fontaine et al. 2015) are shown

Mar Ecol Prog Ser 613: 197–210, 2019

included king penguins, small sized Patagoniantoothfish (TL 43 cm) and the prey item with the low-est δ13C value, adult female southern elephant seals(δ13C = −21.9 ± 0.7‰, n = 70). Group C (δ15N = 12.7 ±0.4‰) contained medium sized toothfish (TL 63 cm),elephant seal pups, Ant arctic fur seals and sub-antarctic fur seals. Group D had the highest δ15N(14.4 ± 0.6‰) and δ13C (−18.9 ± 0.6‰) values andincluded only large (TL 95, 107 and 160 cm) Patagon-ian toothfish. Toothfish of TL 160 cm was the preyitem with the highest δ15N value of all prey items(δ15N = 14.9 ± 0.7‰).

All killer whale isotopic values were inside the95% mixing region of the mixing polygon delimitedby the isotopic values of the 4 prey groups adjustedto DTDFs, thus validating the MixSIAR modelsfitted with these prey groups (Fig. 3b). The bestMixSIAR model (Model 1) included whether or notkiller whales interacted with fisheries before sam-pling (LOOic = 24.2; Table 3). However, this model

and the null model were differentiated by LOOic =0.1, and the 2 models had close weights (0.51 and0.49, res pectively), indicating that the depredationfactor had low explanatory power. According to thenull model, prey Group C was most important in thediet of Crozet killer whales with a mean contribu-tion of 33 ± 19% (Fig. 4a). Group D was the secondmost important prey group with 28 ± 11% meancontribution. Group A and Group B were the leastcontributing prey groups with 17 ± 7 and 22 ± 13%,respectively. According to Model 1, Group C con-tributed 35 ± 21% for depredating samples, and 33± 20% for non-depredating samples (Fig. 4b). Thecontribution of Group D to the diet of depredatingsamples was higher than for non-depredating sam-ples (32 ± 13 and 28 ± 11%, respectively). The dietof non-depredating samples included Group A andB in larger proportions (17 ± 7 and 22 ± 13%,respectively) than for depredating samples (12 ± 7and 21 ± 13%, res pectively).

204

Fig. 2. Isospace of mean (±SD)δ13C and δ15N values of the puta-tive prey items for the Crozetkiller whales (‘KIW_Crozet’). Co -lours indicate prey groups deter-mined by a hierarchical clusteringapproach, and the shape of pointsindicate prey taxa. See Table 2 forspecies associated with the prey

codes used for the figure

Tixier et al.: Toothfish in subantarctic killer whale diet

4. DISCUSSION

The present study provides key insights into therole of toothfish as a prey resource for a generalistsubantarctic predator, the killer whale. The findingsof different stable isotope analysis approaches,which combined inter- and intra-specific compar-isons and dietary reconstruction, suggest that killerwhales do rely on toothfish as well as marine mam-mals and penguins as natural prey items. While fish-eries may facilitate access to toothfish for killerwhales depredating on the fishing gear, fisheries alsoexploit toothfish stocks that are likely used by killerwhales as a natural resource. This study thereforehighlights the trophic interactions between fisheriesand killer whales through depredation and competi-tion for the same resource in subantarctic waters.

Results of the stable isotope analyses confirmedprevious visual observations suggesting that Crozetkiller whales have a generalist feeding strategy.Their niche width was larger than that of specialisedkiller whale populations (Antarctic Type C, B1 andB2; Pitman & Ensor 2003, Krahn et al. 2008, Pitman& Durban 2010, 2012, Durban et al. 2017) butsimilar to that of Marion killer whales, another sub-antarctic population with a generalist feeding strat-egy (Rei singer et al. 2016). The Crozet killer whaleswere sampled in spring/summer and, during thistime of year, the contribution of elephant seal pupsto their diet was higher than that of prey groupsincluding adult elephant seals, penguins and baleenwhales. Recently weaned elephant seal pups are aconcentrated and abundant high-quality food re -source in inshore waters from October to January.

While killer whales may favour elephant sealpups over other prey during that period, whichis consistent with an increase in killer whaleabundance in inshore waters in spring and sum-mer (Guinet 1992), the importance of seals asprey throughout the year remains un known.The fact that the stomach contents of the indi-vidual found dead on Possession Island in win-ter included elephant seal remains suggeststhat this resource may still be consumed duringthat time of year. While this suggestion is sup-ported by skin isotopic values of that dead indi-vidual being similar to values from biopsy sam-

205

Fig. 3. (a) δ13C and δ15N values of the skin samples of Crozet killer whales that depredated or did not depredate toothfish fromfisheries before sampling, and mean with 95% confidence intervals of the putative prey groups (black dots and error bars) es-timated from the source isotopic values and the diet-to-tissue discrimination factors. (b) Mixing polygon including the Crozetkiller whale isotopic values (black dots) and prey groups (mean ± SD isotopic value: white dots and error bars); backgroundshows the probability of prey groups being included in the diet, with probability contours drawn every 10%. Details on species

included in prey groups are provided in Table 2

Model Model LOOic SE ΔLOOic SE Weight# LOOic ΔLOOic

1 Depredation 24.2 8.6 − − 0.512 Null 24.3 9.0 0.1 2.6 0.49

Table 3. MixSIAR model selection outputs based on leave-one-out cross validation information criterion (LOOic). Models werefitted with the occurrence of depredation on toothfish from fish-eries over the 48 d preceding sampling as a fixed effect (‘Depre-dation’), or without any covariate (‘Null’). The LOOic differencesbetween each model and the model with lowest LOOic (ΔLOOic),standard errors (SE) for both LOOic and ΔLOOic values, as well as

the relative weight of models are provided

Mar Ecol Prog Ser 613: 197–210, 2019

ples collected in spring/summer, also suggestinglow seasonal variation in diet, additional samplescollected in winter would be needed to furtherexamine that aspect.

The 2 prey groups (C and D) including mediumand large-sized Patagonian toothfish dominated thediet of Crozet killer whales (over 60% by mass whenpooled, including nearly half from the group exclu-sively made of large toothfish). In addition, there wasno difference in the contribution of these prey groupsbetween depredating and non-depredating samples.The use of isotopic mixing models paired with thefact that none of the prey species included in ouranalysis also prey on toothfish ruled out the possibil-ity of a secondary contamination effect. Instead, thisresult suggests that toothfish may be an importantnatural prey item for Crozet killer whales. This con-clusion is supported by other lines of evidence.Firstly, subantarctic killer whales are able to dive togreat depths (>1000 m at South Georgia; Towers etal. 2019) and they do so when foraging naturally(>750 m at Marion, >300 m at Crozet; Reisinger et al.2015, G. Richard et al. unpubl. data). This depthrange largely overlaps with the bathypelagic depthdistribution of Patagonian toothfish, making thempotentially naturally accessible to killer whales (Ark -hipkin et al. 2003, Collins et al. 2010, Péron et al.2016). Secondly, δ15N values of the Crozet killer

whales, as a proxy of trophic position, were similar tothat of Antarctic Type C killer whales, which areknown to feed preferentially on Antarctic toothfishDissostichus mawsoni (Krahn et al. 2008), a speciesclosely related to the Patagonian toothfish (Collins etal. 2010, Hanchet et al. 2015). Lastly, the isotopicniche of Crozet killer whales partly overlapped withthat of sperm whales from Crozet and Kerguelen,which feed on both Patagonian and Antarctic tooth-fish both naturally (Yukhov 1972) and throughdepredation (Janc et al. 2018, Labadie et al. 2018) inCrozet and Kerguelen waters.

This study therefore suggests that depredation atCrozet is a facilitated behaviour in response to fish-eries making toothfish an aggregated and easilyaccessible resource that killer whales would other-wise naturally forage on, but at higher energeticcosts. By setting their gear at great depths, fisheriesmay provide killer whales with facilitated access tolarge toothfish (>80 cm), which are primarily foundin waters >800 m (Collins et al. 2010). This may ex -plain the greater contribution of that prey group tothe diet of depredating individuals. These findingsare consistent with dietary studies on other de -predating killer whale populations, such as in Gibral-tar Strait, where facilitated access to bluefin tunathrough depredation on fishing lines was found to besubstantially less energetically costly than if this prey

206

Fig. 4. Relative proportions of prey groups in the diet of (a) all Crozet killer whale samples and (b) samples from individualsthat depredated toothfish from fisheries (depredating) or did not (non-depredating) 48 d before sampling. Diet proportionswere estimated from the MixSIAR models (null model and Model 1) and are depicted as boxplots with the median (middleline), 25th and 75th percentiles (box hinges) and 95% confidence intervals (whiskers). Prey groups included the followingitems: Group A: gentoo, macaroni and rockhopper penguins and southern right whales; Group B: king penguins, elephantseals (adult females) and small Patagonian toothfish (TL 43 cm); Group C: southern elephant seals (pups), Antarctic/subantarctic fur seals (pups) and medium Patagonian toothfish (TL 63 cm); Group D: large Patagonian toothfish (TL > 95 cm)

Tixier et al.: Toothfish in subantarctic killer whale diet

species was naturally and actively hunted (Guinet etal. 2007, Esteban et al. 2016).

More broadly, this study presents further evidencethat toothfish have a key role as prey in subantarcticfood web dynamics. While fisheries may facilitateaccess to toothfish for some predators, they may alsoimpact populations that naturally depend on thatresource through direct fish biomass removal. Tooth-fish has been confirmed as natural prey of spermwhales and sleeper sharks Somniosus antarcticus(Yukhov 1972, Cherel & Duhamel 2004) and sug-gested as natural prey for meso- and epi-pelagicpredators such as albatrosses (Cherel et al. 2000,2017). While the commercial exploitation of toothfishstocks is now highly regulated, stocks underwentsubstantial illegal over-exploitation across the South-ern Ocean in the 1990s that likely affected theseapex predator species through direct trophic effects(Kock et al. 2007). The Crozet killer whales under-went a sharp decline in the 1990s and this was partlyattributed to illegal fishers using lethal means torepel whales depredating toothfish (Poncelet et al.2010, Tixier et al. 2015, 2017). However, from ourresults, it is likely that the illegal over-exploitation oftoothfish stocks, paired with substantial decreases ofsouthern elephant seals, king penguins and largewhales (Guinet et al. 1992, Clapham et al. 1999,Weimerskirch et al. 2003, 2018, Pruvost et al. 2015),has also contributed to the decline of this population.Decreased toothfish availability may also havecaused dietary shifts for killer whales in areas wherestocks were depleted. For instance, killer whalesat Marion Island were expected, from observa -tions, to have large isotopic overlap with killerwhales at Crozet (Reisinger et al. 2011), but this wasnot the case. The Marion killer whales are at a lowertrophic level than the Crozet whales, and this dif -ference may be explained by lower toothfish intakebecause tooth fish stocks have been more impactedby illegal fishing at Marion than at Crozet (Boon-zaier et al. 2012) (see Supplement 2 for further discussion).

In summary, this study has provided a preliminaryassessment of the diet of killer whales that consume awide range of subantarctic resources. However, de -termining the diet of a generalist predator is oftenhampered by temporal variations in prey consump-tion and limited information on prey. This is the casefor cephalopods — their contribution to the diet couldnot be assessed in the present study (see Supple-ment 3 for further discussion). Therefore, furtherstudies using higher resolution dietary methods, suchas compound-specific stable isotope or fatty acid

analyses (e.g. Herman et al. 2005, Matthews & Fer-guson 2014), are needed. Despite these limitations,the present study provided new insights on the roleof natural prey preferences in the propensity of killerwhales to switch prey in response to environmentalchanges. Specifically, our results support the as -sump tion that killer whales are more likely to de -velop depredation on fisheries as a new foraging tac-tic if fish is already part of their natural diet. Thisassumption was proposed as explaining why not allkiller whale populations, which greatly differ in preypreferences, switch from natural feeding to depreda-tion despite large overlaps with fishing activity (e.g.Fearnbach et al. 2014, Peterson et al. 2014). The roleof variation in prey preferences at the intra-popula-tion level, as reported in other killer whale popula-tions (e.g. Samarra et al. 2017), should therefore befurther examined with a larger sample size to under-stand the heterogeneity observed across groups/individuals at Crozet in regards to depredation (Tix-ier et al. 2017). More importantly, our findings haveemphasised the importance of toothfish, a species ofhigh commercial value, in the natural diet of killerwhales. However, the amount of toothfish that killerwhales eat naturally compared to the toothfish bio-mass they remove from longlines when depredatingis still unknown. This information is required for (1)assessing the minimum amount of toothfish biomassrequired to sustain killer whale populations, and (2)estimating the extent to which the depredated part ofthat amount (estimated at several hundred t yr−1)should be considered as natural or artificial mortalityin fish stock assessments.

Acknowledgements. The authors thank the fieldworkers ofPossession Island and the fishing vessel crews for their helpin collecting tissue samples. Special thanks are due to Ghis-lain Doremus for retrieving stomach contents, and toMaxime Loubon and Frank Theron for assisting in biopsysampling. We are also grateful to the fishery observers whocollected sighting data from fishing vessels and to theMuseum National d’Histoire Naturelle (Prof. Guy Duhamel,Nicolas Gasco and Charlotte Chazeau) for facilitating accessto these data. We also thank Gael Guillou (LIENSs, LaRochelle), Marina Fontaine and Jeremy Kizska for helpingwith the stable isotope analysis. This work was conducted aspart of the 109 program with the Institut Polaire Francais(IPEV). Funding and logistic support were provided by theTerres Australes et Antarctiques Francaises (TAAF), theReserve Naturelle des Terres Australes, and the ReunionIsland Fisheries Companies (SARPC) as part of the ORCA -DEPRED research program. P.T. and J.P.Y.A. were sup-ported by the Australian Research Council (Linkage Project160100329). R.R.R. and C.G. were supported by the Interna-tional Whaling Commission’s Southern Ocean ResearchPartnership.

207

Mar Ecol Prog Ser 613: 197–210, 2019

LITERATURE CITED

Arkhipkin A, Brickle P, Laptikhovsky V (2003) Variation inthe diet of the Patagonian toothfish with size, depth andseason around the Falkland Islands. J Fish Biol 63: 428−441

Boitani L, Powell RA (2012) Carnivore ecology and conser-vation: a handbook of techniques. Oxford UniversityPress, Oxford

Boonzaier L, Harper S, Zeller D, Pauly D (2012) A brief his-tory of fishing in the Prince Edward Islands, South Africa,1950−2010. In: Harper S, Zylich K, Boonzaier L, Le Man-ach F, Pauly D, Zeller D (eds) Fisheries catch reconstruc-tions: islands, Part II. Fisheries Centre Research Reports,Vol 20. Fisheries Center, University of British Columbia,Vancouver, p 95−101

Capella JJ, Abramson JZ, Vilina YA, Gibbons J (2014)Observations of killer whales (Orcinus orca) in the fjordsof Chilean Patagonia. Polar Biol 37: 1533−1539

Cherel Y, Duhamel G (2004) Antarctic jaws: cephalopodprey of sharks in Kerguelen waters. Deep Sea Res I 51: 17−31

Cherel Y, Hobson KA (2007) Geographical variation in car-bon stable isotope signatures of marine predators: a toolto investigate their foraging areas in the SouthernOcean. Mar Ecol Prog Ser 329: 281−287

Cherel Y, Weimerskirch H, Trouvé C (2000) Food and feed-ing ecology of the neritic-slope forager black-browedalbatross and its relationships with commercial fisheriesin Kerguelen waters. Mar Ecol Prog Ser 207: 183−199

Cherel Y, Fontaine C, Richard P, Labat JP (2010) Isotopicniches and trophic levels of myctophid fishes and theirpredators in the Southern Ocean. Limnol Oceanogr 55: 324−332

Cherel Y, Hobson KA, Guinet C (2015) Milk isotopic valuesdemonstrate that nursing fur seal pups are a full trophiclevel higher than their mothers. Rapid Comm Mass Spec-trom 29(16):1485–1490

Cherel Y, Xavier JC, de Grissac S, Trouvé C, WeimerskirchH (2017) Feeding ecology, isotopic niche, and ingestionof fishery-related items of the wandering albatrossDiomedea exulans at Kerguelen and Crozet Islands. MarEcol Prog Ser 565: 197−215

Clapham PJ, Young SB, Brownell RL Jr (1999) Baleenwhales: conservation issues and the status of the mostendangered populations. Mammal Rev 29: 37−62

Collins MA, Brickle P, Brown J, Belchier M (2010) ThePatagonian toothfish: biology, ecology and fishery. AdvMar Biol 58: 227−300

Constable AJ, de la Mare WK, Agnew DJ, Everson I, MillerD (2000) Managing fisheries to conserve the Antarcticmarine ecosystem: practical implementation of the Con-vention on the Conservation of Antarctic Marine LivingResources (CCAMLR). ICES J Mar Sci 57: 778−791

Croxall JP, Butchart SHM, Lascelles B, Stattersfield AJ, Sul-livan B, Symes A, Taylor P (2012) Seabird conservationstatus, threats and priority actions: a global assessment.Bird Conserv Int 22: 1−34

de Bruyn PJN, Tosh CA, Terauds A (2013) Killer whale eco-types: Is there a global model? Biol Rev Camb Philos Soc88: 62−80

Durban JW, Fearnbach H, Burrows DG, Ylitalo GM, PitmanRL (2017) Morphological and ecological evidence for twosympatric forms of Type B killer whale around theAntarctic Peninsula. Polar Biol 40: 231−236

Esteban R, Verborgh P, Gauffier P, Giménez J, Guinet C, deStephanis R (2016) Dynamics of killer whale, bluefintuna and human fisheries in the Strait of Gibraltar. BiolConserv 194: 31−38

Fearnbach H, Durban JW, Ellifrit DK, Waite JM and others(2014) Spatial and social connectivity of fish-eating ‘Res-ident’ killer whales (Orcinus orca) in the northern NorthPacific. Mar Biol 161: 459−472

Fontaine M, Carravieri A, Simon-Bouhet B, Bustamante Pand others (2015) Ecological tracers and at-sea observa-tions document the foraging ecology of southern long-finned pilot whales (Globicephala melas edwardii ) inKerguelen waters. Mar Biol 162: 207−219

Foote AD, Newton J, Piertney SB, Willerslev E, Gilbert MTP(2009) Ecological, morphological and genetic divergenceof sympatric North Atlantic killer whale populations. MolEcol 18: 5207−5217

Ford JK, Ellis GM, Barrett-Lennard LG, Morton AB, PalmRS, Balcomb KC III (1998) Dietary specialization in twosympatric populations of killer whales (Orcinus orca) incoastal British Columbia and adjacent waters. Can J Zool76: 1456−1471

Gasco N, Tixier P, Duhamel G, Guinet C (2015) Comparisonof two methods to assess fish losses due to depredationby killer whales and sperm whales on demersal long-lines. CCAMLR Sci 22: 1−14

Gilman E, Brothers N, McPherson G, Dalzell P (2007)A review of cetacean interactions with longline gear.J Cetacean Res Manag 8: 215−223

Gilman E, Clarke S, Brothers N, Alfaro-Shigueto J and oth-ers (2008) Shark interactions in pelagic longline fisheries.Mar Policy 32: 1−18

Gilman E, Suuronen P, Hall M, Kennelly S (2013) Causesand methods to estimate cryptic sources of fishing mor-tality. J Fish Biol 83: 766−803

Giménez J, Ramírez F, Almunia J, Forero MG, de StephanisR (2016) From the pool to the sea: applicable isotopeturnover rates and diet to skin discrimination factors forbottlenose dolphins (Tursiops truncatus). J Exp Mar BiolEcol 475: 54−61

Giménez J, Ramírez F, Forero MG, Almunia J, de Stepha-nis R, Navarro J (2017) Lipid effects on isotopic valuesin bottlenose dolphins (Tursiops truncatus) and theirprey with implications for diet assessment. Mar Biol164: 122

Guinet C (1992) Comportement de chasse des orques (Orcinus orca) autour des iles Crozet. Can J Zool 70: 1656−1667

Guinet C, Jouventin P (1990) La vie sociale des “baleinestueuses”. Recherche 220: 508−510

Guinet C, Jouventin P, Weimerskirch H (1992) Populationchanges, movements of southern elephant seals onCrozet and Kerguelen Archipelagos in the last decades.Polar Biol 12: 349−356

Guinet C, Barrett-Lennard LG, Loyer B (2000) Co-ordinatedattack behavior and prey sharing by killer whales atCrozet Archipelago: strategies for feeding on negatively-buoyant prey. Mar Mamm Sci 16: 829−834

Guinet C, Domenici P, de Stephanis R, Barrett-Lennard L,Ford JKB, Verborgh P (2007) Killer whale predation onbluefin tuna: exploring the hypothesis of the endurance-exhaustion technique. Mar Ecol Prog Ser 347: 111−119

Guinet C, Tixier P, Gasco N, Duhamel G (2015) Long-termstudies of Crozet Island killer whales are fundamental tounderstanding the economic and demographic conse-

208

Tixier et al.: Toothfish in subantarctic killer whale diet

quences of their depredation behaviour on the Patagon-ian toothfish fishery. ICES J Mar Sci 72: 1587−1597

Hanchet S, Dunn A, Parker S, Horn P, Stevens D, MormedeS (2015) The Antarctic toothfish (Dissostichus mawsoni): biology, ecology, and life history in the Ross Sea region.Hydrobiologia 761: 397−414

Herman DP, Burrows DG, Wade PR, Durban JW and others(2005) Feeding ecology of eastern North Pacific killerwhales Orcinus orca from fatty acid, stable isotope, andorganochlorine analyses of blubber biopsies. Mar EcolProg Ser 302: 275−291

Jackson AL, Inger R, Parnell AC, Bearhop S (2011) Compar-ing isotopic niche widths among and within communi-ties: SIBER−Stable Isotope Bayesian Ellipses in R. J AnimEcol 80: 595−602

Janc A, Richard G, Guinet C, Arnould JPY and others (2018)How do fishing practices influence sperm whale (Phy-seter macrocephalus) depredation on demersal longlinefisheries? Fish Res 206: 14−26

Jennings S, Kaiser M, Reynolds JD (2009) Marine fisheriesecology. Blackwell Science, Oxford

Knox GA (2006) Biology of the Southern Ocean. CambridgeUniversity Press, Cambridge

Kock KH (2001) The direct influence of fishing and fishery-related activities on non-target species in the SouthernOcean with particular emphasis on longline fishing andits impact on albatrosses and petrels — a review. Rev FishBiol Fish 11: 31−56

Kock KH, Purves MG, Duhamel G (2006) Interactions be -tween cetacean and fisheries in the Southern Ocean.Polar Biol 29: 379−388

Kock KH, Reid K, Croxall J, Nicol S (2007) Fisheries in theSouthern Ocean: an ecosystem approach. Philos Trans RSoc Lond B Biol Sci 362: 2333−2349

Krahn MM, Pitman RL, Burrows DG, Herman DP, PearceRW (2008) Use of chemical tracers to assess diet and persistent organic pollutants in Antarctic Type C killerwhales. Mar Mamm Sci 24: 643−663

Labadie G, Tixier P, Barbraud C, Fay R, Gasco N, DuhamelG, Guinet C (2018) First demographic insights on histor-ically harvested and poorly known male sperm whalepopulations off the Crozet and Kerguelen Islands (South-ern Ocean). Mar Mamm Sci 34: 595−615

Laws RM (1977) Seals and whales of the Southern Ocean.Philos Trans R Soc Lond B Biol Sci 279: 81−96

Lesage V, Morin Y, Rioux È, Pomerleau C, Ferguson SH,Pelletier É (2010) Stable isotopes and trace elements asindicators of diet and habitat use in cetaceans: predictingerrors related to preservation, lipid extraction, and lipidnormalization. Mar Ecol Prog Ser 419: 249−265

Martin A, Pruvost P (2007) Pecheker, relational database foranalysis and management of halieutic and biologicaldata from the scientific survey of the TAAF fisheries.Muséum National d’Histoire Naturelle, Paris. https: //borea. mnhn.fr/pecheker

Matthews CJ, Ferguson SH (2014) Spatial segregation andsimilar trophic-level diet among eastern Canadian Arctic/ north-west Atlantic killer whales inferred frombulk and compound specific isotopic analysis. J Mar BiolAssoc UK 94: 1343−1355

Mitchell JD, McLean DL, Collin SP, Langlois TJ (2018)Shark depredation in commercial and recreational fish-eries. Rev Fish Biol Fish 28: 715−748

Newsome TM, Dellinger JA, Pavey CR, Ripple WJ, ShoresCR, Wirsing AJ, Dickman CR (2015) The ecological

effects of providing resource subsidies to predators. GlobEcol Biogeogr 24: 1−11

Northridge SP, Hofman RJ (1999) Marine mammal inter -actions with fisheries. In: Twiss JR, Reeves RR (eds) Con-servation and management of marine mammals. Smith-sonian Press, Washington, DC, p 99−119

Péron C, Welsford DC, Ziegler P, Lamb TD and others (2016)Modelling spatial distribution of Patagonian toothfishthrough life-stages and sex and its implications for thefishery on the Kerguelen Plateau. Prog Oceanogr 141: 81−95

Peterson MJ, Mueter F, Criddle K, Haynie AC (2014) Killerwhale depredation and associated costs to Alaskansablefish, Pacific halibut and Greenland turbot longlin-ers. PLOS ONE 9: e88906

Pitman RL, Durban JW (2010) Killer whale predation onpenguins in Antarctica. Polar Biol 33: 1589−1594

Pitman RL, Durban JW (2012) Cooperative hunting behav-ior, prey selectivity and prey handling by pack ice killerwhales (Orcinus orca), type B, in Antarctic Peninsulawaters. Mar Mamm Sci 28: 16−36

Pitman RL, Ensor P (2003) Three forms of killer whales(Orcinus orca) in Antarctic waters. J Cetacean ResManag 5: 131−140

Poncelet É, Barbraud C, Guinet C (2010) Population dynam-ics of killer whales (Orcinus orca) in the Crozet Archi -pelago, southern Indian Ocean: a mark−recapture studyfrom 1977 to 2002. J Cetacean Res Manag 11: 41−48

Pruvost P, Duhamel G, Gasco N, Palomares MD (2015) Ashort history of the fisheries of Crozet Islands. In: Palo-mares MLD, Pauly D (eds) Marine fisheries catches ofsubantarctic islands, 1950-2010. Fisheries Centre Re -search Reports, Vol 23. Fisheries Centre, University ofBritish Columbia, Vancouver, p 31−37

R Development Core Team (2017) R: a language and envi-ronment for statistical computing. R Foundation for Sta-tistical Computing, Vienna

Read AJ (2008) The looming crisis: interactions betweenmarine mammals and fisheries. J Mammal 89: 541−548

Reisinger RR, de Bruyn PJN, Tosh CA, Oosthuizen WC,Mufanadzo NT, Bester MN (2011) Prey and seasonalabundance of killer whales at sub-Antarctic MarionIsland. Afr J Mar Sci 33: 99−105

Reisinger RR, Keith M, Andrews RD, de Bruyn PJN (2015)Movement and diving of killer whales (Orcinus orca) at aSouthern Ocean archipelago. J Exp Mar Biol Ecol 473: 90−102

Reisinger RR, Gröcke DR, Lübcker N, McClymont EL,Hoelzel AR, de Bruyn PJN (2016) Variation in the diet ofkiller whales Orcinus orca at Marion Island, SouthernOcean. Mar Ecol Prog Ser 549: 263−274

Reisinger RR, Raymond B, Hindell MA, Bester MN and oth-ers (2018) Habitat modelling of tracking data from multi-ple marine predators identifies important areas in theSouthern Indian Ocean. Divers Distrib 24: 535−550

Roche C, Guinet C, Gasco N, Duhamel G (2007) Marinemammals and demersal longline fishery interactions inCrozet and Kerguelen Exclusive Economic Zones: anassessment of depredation levels. CCAMLR Sci 14: 67−82

Samarra FP, Vighi M, Aguilar A, Víkingsson GA (2017)Intra-population variation in isotopic niche in herring-eating killer whales off Iceland. Mar Ecol Prog Ser 564: 199−210

Sidorovich VE, Tikhomirova LL, Jedrzejewska B (2003) WolfCanis lupus numbers, diet and damage to livestock in

209

Mar Ecol Prog Ser 613: 197–210, 2019

relation to hunting and ungulate abundance in north-eastern Belarus during 1990−2000. Wildl Biol 9: 103−111

Sillero-Zubiri C, Sukumar R, Treves A (2007) Living withwildlife: the roots of conflict and the solutions. In: Mac-Donald D, Service K (eds) Key topics in conservationbiology. Oxford Press, Oxford, p 266−272

Similä T, Holst JC, Christensen I (1996) Occurrence and dietof killer whales in northern Norway: seasonal patternsrelative to the distribution and abundance of Norwe-gian spring-spawning herring. Can J Fish Aquat Sci 53: 769−779

Smith JA, Mazumder D, Suthers IM, Taylor MD (2013) To fitor not to fit: evaluating stable isotope mixing modelsusing simulated mixing polygons. Methods Ecol Evol 4: 612−618

Stock BC, Semmens BX (2013) MixSIAR GUI user manual,version 1.0. http: //conserver.iugo-cafe.org/user/brice.semmens/ MixSIAR

Stock BC, Jackson AL, Ward EJ, Parnell AC, Phillips DL,Semmens BX (2018) Analyzing mixing systems using anew generation of Bayesian tracer mixing models. PeerJ6: e5096

Stoddart LC, Griffiths RE, Knowlton FF (2001) Coyote re -sponses to changing jackrabbit abundance affect sheeppredation. J Range Manage 54: 15−20

Tasker ML, Camphuysen CJ, Cooper J, Garthe S, Mon-tevecchi WA, Blaber SJ (2000) The impacts of fishing onmarine birds. ICES J Mar Sci 57: 531−547

Tixier P, Gasco N, Duhamel G, Viviant M, Authier M, GuinetC (2010) Interactions of Patagonian toothfish fisherieswith killer and sperm whales in the Crozet Islands Exclu-sive Economic Zone: an assessment of depredation levelsand insights on possible mitigation strategies. CCAMLRSci 17: 179−195

Tixier P, Gasco N, Guinet C (2014) Killer whales of theCrozet Islands: photo-identification catalogue 2014.Centre d’Etudes Biologiques de Chizé-CNRS, Villiersen Bois

Tixier P, Authier M, Gasco N, Guinet C (2015) Influence ofartificial food provisioning from fisheries on killer whalereproductive output. Anim Conserv 18: 207−218

Tixier P, Gasco N, Duhamel G, Guinet C (2016) Depredationof Patagonian toothfish (Dissostichus eleginoides) by twosympatrically occurring killer whale (Orcinus orca) eco-

types: insights on the behavior of the rarely observedtype D killer whales. Mar Mamm Sci 32: 983−1003

Tixier P, Barbraud C, Pardo D, Gasco N, Duhamel G, GuinetC (2017) Demographic consequences of fisheries interac-tion within a killer whale (Orcinus orca) population. MarBiol 164: 170

Torres LG, Rayment W, Olavarría C, Thompson DR and oth-ers (2017) Demography and ecology of southern rightwhales Eubalaena australis wintering at sub-AntarcticCampbell Island, New Zealand. Polar Biol 40: 95−106

Towers JR, Tixier P, Ross KA, Bennett J, Arnould JP, PitmanRL, Durban JW (2019) Movements and dive behaviour ofa toothfish-depredating killer and sperm whale. ICES JMar Sci 76: 298−311

Travers T, van den Hoff J, Lea MA, Carlyon K, Reisinger R,de Bruyn PN, Morrice M (2018) Aspects of the ecology ofkiller whale (Orcinus orca Linn.) groups in the near-shore waters of Sub-Antarctic Macquarie Island. PolarBiol 41: 2249−2259

Van Valen L (1965) Morphological variation and width ofecological niche. Am Nat 99: 377−390

Vehtari A, Gelman A, Gabry J (2017) Practical Bayesianmodel evaluation using leave-one-out cross-validationand WAIC. Stat Comput 27: 1413−1432

Votier SC, Furness RW, Bearhop S, Crane JE and others(2004) Changes in fisheries discard rates and seabirdcommunities. Nature 427: 727−730

Weimerskirch H, Inchausti P, Guinet C, Barbraud C (2003)Trends in bird and seal populations as indicators of a sys-tem shift in the Southern Ocean. Antarct Sci 15: 249−256

Weimerskirch H, Le Bouard F, Ryan PG, Bost CA (2018)Massive decline of the world’s largest king penguincolony at Ile aux Cochons, Crozet. Antarct Sci 30: 236−242

Woodroffe R, Thirgood S, Rabinowitz A (2005) People andwildlife: Conflict or co-existence? Cambridge UniversityPress, Cambridge

Yukhov VL (1972) The range of fish of the genus Dissos-tichus (Fam. Nototheniidae) in Antarctic waters of theIndian Ocean. J Ichthyol 12: 346−347

Yurkowski DJ, Hussey NE, Semeniuk C, Ferguson SH, FiskAT (2015) Effects of lipid extraction and the utility of lipidnormalization models on δ13C and δ15N values in Arcticmarine mammal tissues. Polar Biol 38: 131−143

210

Editorial responsibility: Keith Hobson, London, Ontario, Canada

Submitted: October 9, 2018; Accepted: February 12, 2019Proofs received from author(s): March 14, 2018